Using a PPI column in PI

Data Modeling

Using a PPI column in PI

Hi All,

I have a table that has PI consisting of Col_A and Col_B. Col_C is a timestamp(6) field which is a PPI.

As per the current data, the table has a skew factor of 75%.

Therefore, I created an identical table, this time with PI consisting of Col_A, Col_B and Col_C, with Col_C again as PPI.

Now, when I loaded the records that are in the existing table to this new one, the skew factor is around 5%.

However, I am not sure if this has any other implications in terms of data processing times or anything else.

Please advise.



Tags (4)

Re: Using a PPI column in PI

Hi Aarsh,

any way 75% skew is not acceptable. So good you created PI with all three column and it is now only 5%.

Now as per your usage of the SQL on this table there are possible scenarios -

If you able to use all three column in where clause, you are going to get one AMP operation. (this is first priority)

If it is not possible then ensure to use col_c in where clause which does partision elimination and it is good.

If you don't use atleaset col_c in in where clause then I prefer remove PPI live with only PI on three columns.

So that many of queries would go for all AMP operation and it is fine.



Teradata Employee

Re: Using a PPI column in PI

Can you choose a better PI [single or combination of columns] excluding timestamp column?